Ultrasound imaging system with a neural network for deriving imaging data and tissue information

Abstract

An ultrasound system according to some embodiments may include an ultrasound transducer configured to transmit ultrasound pulses toward tissue and generate echo signals responsive to the ultrasound pulses, a channel memory configured to store the echo signals, a beamformer configured to generated beamformed signals responsive to the echo signals, a neural network configured to receive one or more samples of the echo signals or the beamformed signals and produce a first type of ultrasound imaging data, and a processor configured to generate a second type of ultrasound imaging data, wherein the one or more processors may be further configured to generate an ultrasound image based on the first type of ultrasound imaging data and the second type of ultrasound imaging data and to cause a display communicatively coupled therewith to display the ultrasound image.

What is claimed is: 
     
       1. An ultrasound system comprising:
 an ultrasound transducer configured to transmit ultrasound pulses toward tissue and generate echo signals responsive to the ultrasound pulses; 
 a channel memory configured to store the echo signals; 
 a beamformer configured to generate beamformed radiofrequency (RF) signals responsive to the echo signals; 
 a graphics processing unit (GPU) configured to receive input data comprising one or more samples of the echo signals or the beamformed RF, wherein the one or more samples represent less than all of the echo signals or the beamformed RF signals, and the GPU is further configured to execute instructions to perform a machine-trained algorithm to generate a first type of ultrasound imaging data based on the input data, wherein the machine-trained algorithm is configured to select an operational mode for processing the input data from a plurality of operational modes responsive to user input or automatically set by the ultrasound system based on an imaging mode of the ultrasound system during acquisition of the echo signals; and 
 a processor configured to generate a second type of ultrasound imaging data based on the beamformed RF signals, wherein the processor is further configured to generate an ultrasound image based on the first type of ultrasound imaging data and the second type of ultrasound imaging data. 
 
     
     
       2. The ultrasound imaging system of  claim 1 , wherein the second type of ultrasound imaging data comprises B-mode imaging data, and wherein the first type of ultrasound imaging data comprises one of Doppler imaging data, vector flow imaging data, elastography imaging data, tissue type characterization data, wall shear stress of an anatomical structure containing a fluid therein, tissue composition data, ultrasound contrast agent information, plaque characterization data, one or more diagnostic indicators associated with the B-mode imaging data, or combinations thereof. 
     
     
       3. The ultrasound system of  claim 1 , wherein the machine-trained algorithm comprises a neural network. 
     
     
       4. The ultrasound system of  claim 3 , wherein the neural network includes a deep neural network (DNN) or a convolutional neural network (CNN). 
     
     
       5. The ultrasound imaging system of  claim 1 , wherein the GPU and the transducer are disposed within an ultrasound probe. 
     
     
       6. The ultrasound imaging system of  claim 1 , wherein the GPU is disposed in a computing device which is separate from an ultrasound probe. 
     
     
       7. The ultrasound imaging system of  claim 1 , further comprising a data selector configured to select the one or more samples of the echo signals or the beamformed RF signals as the input data. 
     
     
       8. The ultrasound imaging system of  claim 7 , wherein the data selector is further configured to selectively couple one of the samples of echo signals or the samples of the beamformed RF signals to the machine-trained algorithm responsive to a control signal received by the data selector. 
     
     
       9. The ultrasound system of  claim 8 , wherein the control signal is generated responsive to user input. 
     
     
       10. The ultrasound imaging system of  claim 1 , wherein the processor is further configured to cause a display to display the ultrasound image. 
     
     
       11. The ultrasound imaging system of  claim 1 , wherein the machine-trained algorithm is further configured to receive auxiliary data as input, the auxiliary data including ultrasound transducer configuration information, beamformer configuration information, information about the tissue being imaged, or combinations thereof, and wherein the first type of ultrasound imaging data provided by the machine-trained algorithm is further based on the auxiliary data. 
     
     
       12. The ultrasound imaging system of  claim 1 , wherein the machine-trained algorithm is operatively associated with a training algorithm configured to receive an array of training inputs and known outputs, wherein the training inputs comprise echo signals, beamformed signals, or combinations thereof associated with a region of imaged tissue and the known outputs comprise known properties of the imaged tissue. 
     
     
       13. The ultrasound imaging system of  claim 12 , wherein the known properties comprise properties of images obtained using an imaging modality other than ultrasound. 
     
     
       14. The ultrasound system of  claim 1 , wherein the machine-trained algorithm is configured to predict a fat content of the tissue based on the input data without use of the second type of ultrasound imaging data. 
     
     
       15. The ultrasound system of  claim 1 , wherein the machine-trained algorithm is configured to predict flow properties of a fluid contained in an anatomical structure of the tissue based on temporally successive samples of the input data without the use of quadrature signals produced by an image processing circuit. 
     
     
       16. The ultrasound system of  claim 1 , wherein the machine-trained algorithm is configured to produce predicted beamformed signals based on samples of the echo signals, and to use the predicted beamformed signals to generate the first type of ultrasound imaging data.